Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
* thomz joins02:42
* ajs6f joins08:54
* peichman joins09:02
* arebenji joins09:08
* thomz leaves09:14
* peichman leaves09:15
* acoburn joins09:23
* apb18 joins09:33
* dchandekstark joins09:42
awoods: so this delete query that Jim Coble posted about is really bugging me09:43
i'm trying to track where it's coming from and see if it's really necessary09:44
<acoburn>dchandekstark: FWIW, it bugs me, too
<dchandekstark>i see where it's coming from in ispn09:45
seems to be here: https://github.com/infinispan/infinispan/blob/7.2.4.Final/persistence/jdbc/src/main/java/org/infinispan/persistence/jdbc/stringbased/JdbcStringBasedStore.java#L278
<acoburn>dchandekstark: can't you just add an index to that table?
a full table scan is a terrible waste of resources09:46
<dchandekstark>acoburn: that was the question Jim raised, I am prepared to try that
but still, i think it's worth determining if this query is even necessary
<awoods>dchandekstark: Are you looking for help or validation?09:47
<dchandekstark>something is calling purge() on the store periodically
awoods: both, really :)
This issue is causing some acute pain09:48
<awoods>dchandekstark: Do you see a risk in setting up the index as mentioned in the email thread?09:49
<dchandekstark>many write operations are failing due to various timeouts b/c blocked behind the table lock of this delete query
awoods: not really, but it's curious that ispn doesn't do it09:50
maybe it's a bug ;)09:51
<awoods>dchandekstark: As for the period purge(), that is likely ModeShape housekeeping. As a note, ModeShape5 does completely away with ISPN.
<dchandekstark>awoods: ok, i can look at modeshape - and yeah, i knew about MS dropping ispn09:53
which fcrepo 4.6 will use, right?
awoods: is there a rough target date for that rel?09:54
<awoods>dchandekstark: Actually, it is timely that you ask.
<dchandekstark>awoods acoburn - we're going to try indexing the version col fwiw
<awoods>dchandekstark: The primary blocker for releasing a ModeShape5-based Fedora is the lack of testing and tooling around migrating from a ModeShape4 backend to a ModeShape5 backend.09:56
<dchandekstark>awoods: that's pretty big
<awoods>dchandekstark: There is currently /fcr:backup and /fcr:restore which could be used for this purpose. Basically those endpoints expose the recommended migration approach suggested by the ModeShape project.
dchandekstark: escowles has offered to test those endpoints for a test Princeton migration to ModeShape5.09:57
<dchandekstark>awoods: given that we just spent about 6 months migrating to fcrepo4, i think we're not exactly enthusiastic about another "migation" :)
<awoods>dchandekstark: That is completely understandable.09:58
<dchandekstark>awoods: btw, we finished our migration right on schedule - end of FY15/16 - :)
<awoods>dchandekstark: congrats!09:59
dchandekstark: You and coblej are a force.
<dchandekstark>awoods: we're something anyway!10:00
<awoods>dchandekstark: The other migration option, half of which exists, is working directly against the Fedora API with tooling that export Fedora RDF and optionally binaries (exists) to the filesystem, then importing those same artifacts into a new Fedora repository via the Fedora API (not exist). It is this import/export that will be the focus of the sprint that is currently in planning.10:01
<dchandekstark>awoods: given some of the issues we've encountered i wouldn't be surprised to find that backup/restore fails10:02
that is to say is not error-free
at least with a repository the size of ours
awoods: the export/import option feels more appealing to me10:03
<awoods>dchandekstark: You are probably right. It would be great to get your engagement in the import/export effort... as a testing stakeholder, if nothing else.
<dchandekstark>as with the migration from 3 to 4, doing it so to speak atomically, i think was a good thing
both for the migration itself as well as the custom validation process we implemented to check behind the migration10:04
<awoods>dchandekstark: And by "atomically", you mean object by object?
<dchandekstark>awoods: i meant to reply to your email that we (I) can certainly be engaged at some level
and want to
awoods: re atomically, basically yes10:05
awoods: at a higher level, we migrated in chunks - i.e., collections
which was, i think, really useful for making adjustments along the way as issues came up10:06
<awoods>dchandekstark: That is good feedback from the field.
<ajs6f>import/export++
<dchandekstark>awoods: well, we knew we didn't want to get x% thru it and have to start over10:07
this way, we could do one collection, verify that it was sucessful, rinse, repeat
awoods acoburn - we're doing this in our dev and test repo dbs: create index version_idx on ispn_entry_FedoraRepository (version);10:08
<awoods>dchandekstark: It will be very helpful to hear if that helps. If not, the next step is likely to dig into the ModeShape/ISPN interaction.10:09
<acoburn>dchandekstark++ # it would be great to hear your results on the tech email list
<dchandekstark>awoods acoburn - we'll report there10:10
* arebenji leaves10:16
* ajs6f leaves10:51
* dchandekstark leaves10:59
* acoburn leaves12:23
* acoburn joins12:24
* bseeger joins12:42
<awoods>afk13:14
* peichman joins13:15
* arebenji joins13:57
* dchandekstark joins14:00
awoods acoburn - initial results from adding the column index are promising, or at least not discouraging14:01
the real test will be in production :O
* mohamedar joins14:08
* peichman leaves14:11
* bseeger leaves14:12
<mohamedar>@acoburn: I have closed PR#92 and created PR#93: https://github.com/fcrepo4-exts/fcrepo-camel-toolbox/pull/93 with the new implementation for https://jira.duraspace.org/browse/FCREPO-206114:13
<acoburn>mohamedar: thanks so much! I am probably not going to be able to look at it closely until the end of next week, but it is on my radar14:16
* peichman joins14:38
<mohamedar>acoburn: Thanks!14:44
* bseeger joins14:56
* f4jenkins leaves15:41
* f4jenkins joins15:42
* jackhill leaves
* jackhill joins15:44
* ajs6f joins15:56
<acoburn>awoods: I can work on the fcrepo-1960 followon stuff either late next week or the following week (unless someone beats me to it)16:26
<awoods>acoburn: Let me see if I can through a PR against it in the meantime.16:28
acoburn: s/through/throw/
<acoburn>awoods++
awoods: I could cut a release of fcrepo-java-client either this evening or over the weekend. I certainly don't want to hold things up16:39
<awoods>acoburn: I am more curious about how bpennell's commit plays into that.
acoburn: how does his commit relate releasing 4.6.0?16:40
acoburn: how does his commit relate to releasing 4.6.0?
<acoburn>awoods: it has nothing to do with 4.6.0
awoods: I mean 0.1.3
awoods: I have code that depends on fcrepo-java-client (camel-toolbox, fcrepo-camel)16:41
<awoods>acoburn: ooohhh... you were talking about cutting a fcrepo-java-client release. got it.
<acoburn>awoods: both of which will require changes when the java-client api changes
awoods: yes, and I want the most recent commits in that release
<awoods>acoburn: if you could cut a release sooner, that would be great... because it seems like bpennell's work is getting close to ready.16:42
<acoburn>awoods: I'll try to do it this afternoon then
<awoods>acoburn++
* dchandekstark leaves16:44
<acoburn>awoods: cross your fingers — a thunderstorm is starting…16:45
<awoods>acoburn: type fast16:46
<ajs6f>awoods: I have totally lost track of where bpennell got to, but I recall thinking "This isn't totally perfect, but it is definitely a big step in the right direction." IOW, I wouldn't block it post-acoburn's release.
<awoods>ajs6f: do you have time to give the current state of bpennell's PR a re-review?16:47
<ajs6f>awoods: no, i also have a tuhnderstorm. A 2 year old thunderstorm who just invented the game of yelling "Boo!" and dumping his glass of water on the floor. Maybe early next week? Will that help?16:48
<awoods>ajs6f: of course... anything is better than nothing.
Boo!
<ajs6f>awoods: Thank you, awoods. I will check your diaper later.16:49
awoods: You may also enjoy the other game he invented of wrapping a diaper around each leg and running around the house screaming "Bye-bye!"16:50
* github-ff joins
[fcrepo-java-client] acoburn tagged fcrepo-java-client-0.1.3 at e7aa314: https://git.io/vKe53
fcrepo-java-client/fcrepo-java-client-0.1.3 2f5ad44 Aaron Coburn: [maven-release-plugin] prepare release fcrepo-java-client-0.1.3
* github-ff leaves
<awoods>ajs6f: I am always looking for new ideas, thanks!16:51
* travis-ci joins16:53
fcrepo4-exts/fcrepo-java-client#38 (fcrepo-java-client-0.1.3 - 2f5ad44 : Aaron Coburn): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-java-client/commit/2f5ad44a5604
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-java-client/builds/141733586
* travis-ci leaves
* github-ff joins17:00
[fcrepo-java-client] acoburn pushed 1 new commit to gh-pages: https://git.io/vKedM
fcrepo-java-client/gh-pages 0a09b38 Aaron Coburn: Creating site for fcrepo-java-client, 0.1.3
* github-ff leaves
* apb18 leaves17:03
<acoburn>awoods: the release is done17:04
<awoods>acoburn: that was quick
<acoburn>awoods: you said "type fast"
<awoods>acoburn: one blocker down, thanks.
<acoburn>awoods: I'll announce it next week
awoods: it's good to wait for maven central to have the artifacts17:05
<awoods>acoburn: great. Are you offline early next week, or just busy?
<acoburn>awoods: busy
* travis-ci joins17:06
fcrepo4-exts/fcrepo-java-client#39 (master - 2606aaf : Aaron Coburn): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-java-client/compare/08565d93b8b9...2606aafdd778
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-java-client/builds/141736264
* travis-ci leaves
* peichman leaves
* bseeger leaves17:07
<ajs6f>awoods: jcoble's long-running SQl query—you don't think that could relate in any way to background binary GC?17:15
awoods: I don't see how, but I'm just throwing spaghetti at the wall.
<awoods>ajs6f: I thought about that ...but ModeShape does binary GC every 24 hours, by default.17:16
awoods: ...but maybe.
<ajs6f>awoods: well, now there's wet noodles all over the floor, so watch where you walk.
<awoods>acoburn: can you update this: https://wiki.duraspace.org/display/FF/Component+Compatibility+Matrix17:18
* acoburn leaves17:19
* ajs6f leaves18:01
* mohamedar1 joins
* mohamedar leaves18:04
* mohamedar1 leaves18:06
* awoods leaves18:07
* awoods joins18:08
* github-ff joins19:28
[fcrepo-camel-toolbox] acoburn opened pull request #94: Update java client version (master...fcrepo-2068) https://git.io/vKvIq
* github-ff leaves
* mikeAtUVa joins21:34
* mikeAtUVa leaves22:30
* f4jenkins joins23:11
* arebenji leaves23:53

Generated by Sualtam